JPS63126075A - Image processor - Google Patents

Abstract

PURPOSE:To execute proper filter processing to a feature to be emphasized by using a filter factor determined by the brightness of peripheral image around a picture element. CONSTITUTION:The data consisting of mXm picture elements around a picture element (i, j) are transferred from an original image memory 1 to a 1st picture element segmenting device 3 by an address counter 2 and the brightness of the image consisting of mXm picture elements is added to the data by an adder 5. The added value is stored in a filter factor storing memory 6. On the other hand, data consisting of 3X3 picture elements around the (i, j) are sent from the memory 1 to a 2nd picture element segmenting device 4. The data are transferred to a product summing device 7 by the 2nd switch 4 and the memory 6. The product summing device 7 executes multiplication and addition of the filter factor selected by the added result and the data of 3X3 picture elements and the calculated result is stored in an computed result storing memory 8 on the basis of an address applied from the address counter 2.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image processing apparatus that performs image processing such as emphasizing and extracting characteristic parts of an image input from a TV camera.

(Prior art) Conventionally, in image processing, spatial filter processing has been applied to images input from a TV camera to emphasize only the characteristic parts in order to recognize the characteristic parts.
However, when the input image shows multiple different objects or when comparing completely different backgrounds to the objects, it is difficult to clarify only the characteristic parts with the same filter processing. There was something I cannot stress enough. As a method for solving this fL problem, Japanese Patent Application Laid-Open No. 61-201372 proposes a method in which the filter coefficients are changed to perform filter processing suitable for the input image. However, in this method, when changing the filter coefficient for each pixel, it is necessary to know in advance what kind of image will be input.

(Problems to be Solved by the Invention) The present invention solves the problem that when a filter process with fixed coefficients is applied to all pixels of an image as in the conventional method, the surface condition of the copy may be different or the same. When there is a large difference in brightness in a copy or the background of a noteworthy copy is different, noise may appear in addition to the expected parts, and other parts are emphasized. The present invention solves the problem of many hot images, and in order to be able to emphasize only the expected features of the image, no matter what kind of image is input, each pixel undergoing spatial filter processing is An object of the present invention is to provide an image processing device that can perform the following steps while determining filter coefficients.

(Means for Solving the Problems) This ninth aspect of the present invention provides an original image memory capable of filter processing, and a first image serving as precise data for determining filter coefficients. a first pixel cutter that cuts out data from the image memory; an adder that adds the data cut out by the first pixel cutter; and an adder that inputs the addition result of the adder as an address and stores a plurality of filter coefficients. a filter coefficient storage memory, a second pixel cutter for cutting out a second image for filter processing from the original image memory; and a second pixel cutter for cutting out a second image for filter processing from the original image memory; An image processing device comprising: a product-summer that performs a filter operation between the second images outputted from the output device; and a calculation result storage memory that stores the output result of the product-summer. It's a friend.

(Example) As shown in FIG. 3, for example, when a general spatial filter performs filter processing of 3 x 3 pixels on an image of 512 pixels x 512 pixels, the calculated value d'(1, j) is expressed by equation (1).

al (4,,1)=d(i-x,j-1)・A+d(
soil + j-x)・B+d(1+1.j-1)・C+d(1
-1,j )・p+a(1,j)・B+a(i+1.
j )・F+ a(i-x,,1+t)・G+d(1
, j+1)・H+d(1+1.j+1)・工・・・(
1] Here, A, B, C, D, E, F, G, H, ■ are fixed filter coefficients. 1. j is 512 x 51
This is the address corresponding to pixel 2, and d(1,5) is (
i, j) means the brightness at a point.

In the present invention, as shown in FIG. 2, the filter coefficients are different from fixed filter coefficients: 1 coefficient Fl, F2 . F3
゜F4. F5. F6. F7. Set it to F8+F9. F1
~F9 is represented by formulas (2) ~ (1G).

F2 = f2 () ... (3) F3 = j'a
() ... (4) F4 = J"4 () ・
... (5) Fs = j'5 () ... (6) F6
=j°6() ...(7]F7 =j'r
() ... (8)F8 =j's (
) ・・・(9)F9=j'9() ・・・
(This is the sum of the brightness of the pixel 11 of mxm pixels centered on the IC1. As a result, the filter processing result obtained by the present invention is 2qυ.

a'(i, j) = d(1-1, j-1)Fll +d
(1,j-1)・F2+d(1+i,,1-1),F3
+a(i-x,j)・F4+d(1,j)・F5+d(
1+L, j)・F6+ d(i-1, j+1)φ
77+cl(1,j+1)・pg+a(4+t,,1+
x) @Fg...α〃 In this way, the filter coefficient is determined by the surrounding image condition (brightness) centered on the pixel for which filter processing is to be applied, so it is necessary to select an appropriate filter coefficient for the feature that is not to be emphasized. It becomes possible to apply a filter.

FIG. 1 is a block diagram of an image processing device that is an embodiment of the present invention. The original image memory includes an original image memory 1 that stores the brightness of each pixel, and an address counter 2 that generates a memory address. .

Original image memory 11 = u(i,
The data of mXm pixels (m is a positive integer) centered at j) is transferred to the first pixel extractor 3, and the adder 5 adds the brightness of the mXm pixel image. This addition result is given as an address for selecting the contents of the filter coefficient storage memory 6. The filter coefficients selected in the filter coefficient storage memory 6 are F as shown in equations (2) to (2).
1 to F9. - From the power source image memory 1, (1, j
) 'r-centered 3×3 pixel data is also sent to the second pixel cutter 4. In the embodiment, the number of pixels is 3×3 pixels, but other numbers such as 5×5.7×7 pixels may be used. From the second pixel extractor 4 and filter coefficient storage memory 6,
The data is transferred to the product and fli7, and the product-summer 7 multiplies and adds the filter coefficient selected based on the addition result by the adder 5 and the data of the 3×3 pixels, and performs the filter processing shown by the formula C1υ. is applied, and this result is stored in the calculation result storage memory 8 based on the address given by the address counter 2.

(Effects of the Invention) According to the present invention, filter coefficients can be selected depending on the brightness characteristics of the original image, so that no matter what image is input, appropriate filter processing can be performed. This provides an image processing device that can emphasize only the desired features.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an image processing device that is an embodiment of the present invention, FIG. 2 is an explanatory diagram showing an overview of filter processing in the device shown in FIG. 1, and FIG. 3 is an overview of filter processing in a conventional device. An explanatory diagram of 0 1...Original image memory 2...Address counter 3
...mXm pixel (first pixel) cutter 4...3X3
Pixel (second pixel) extractor 5... adder 6
... Filter coefficient storage memory 7 ... Multiplier-summer 8 ... Calculation result storage memory

Claims (1)

[Claims] Original image memory that can be filtered,
a first pixel cutter that cuts out a first image serving as data for determining a filter coefficient from the original image memory; an adder that adds the data cut out by the first pixel cutter; a filter coefficient storage memory that inputs the addition result of the adder as an address and stores a plurality of filter coefficients; a second pixel cutter that cuts out a second image for filter processing from the original image memory; and the filter. A product-sum unit that performs a filter operation between the filter coefficients output from the coefficient storage memory and the second image output from the second pixel extractor, and stores the output results of the product-sum unit. An image processing device comprising: a calculation result storage memory.